Authors: Swaminathan Mani
Abstract: Generative Artificial Intelligence has changed the way we make images from text. Now we can make quality images from what we write. This is because of models that use special architectures. These models are really good at making images that look real. Are about the right thing. We can use these models to make art, design and ads. They are also useful in education, healthcare and gaming. This saves us time and money because we do not have to make images by hand. This paper is about how we can make images from text using Generative Artificial Intelligence. We look at how the models work and what's new about them. We talk about models like Stable Diffusion, DALL·E and Imagen. We look at how the whole process works, from getting the text ready to making the image. We also think about how to make the images better by using the words. We discuss what is good and bad about the models we have now. We also look at what other people have found out about making images from text. We compare the ways to do it and talk about what is new and interesting. We think about how we can make images that're just what we want and how we can make the models work better and faster. Generative Artificial Intelligence models that use diffusion are good at making images that look real. Are about the right thing. They open up possibilities, for art and industry. This paper ends by talking about what we need to do to make the models better and more responsible.
